Update README.md

README.md

@@ -21,7 +21,7 @@ tags:
 ## Introduction:
  <p>
   <strong>LongVILA-R1-7B</strong> supports both <u>multiple-choice</u> questions and <u>open-ended</u> questions. It can switch between thinking and non-thinking modes.<br> 
-  <strong>LongVILA-R1-7B</strong> demonstrates strong performance in long video reasoning, achieving <strong>70.7%</strong> on VideoMME (w/ sub.) and surpassing Gemini-1.5-Pro across diverse reasoning tasks.<br> 
+  <strong>LongVILA-R1-7B</strong> demonstrates strong performance in long video reasoning, achieving <strong>71.1%</strong> on VideoMME (w/ sub.) and surpassing Gemini-1.5-Pro across diverse reasoning tasks.<br> 
   <strong>LongVILA-R1-7B</strong> supports processing up to <strong>8,192</strong> video frames per video, with configurable FPS settings.<br>
   <strong>Long-RL</strong> is a codebase that accelerates long video RL training by up to <strong>2.1×</strong> through its MR-SP system. It supports RL training on image, video, and omni inputs across VILA, Qwen/Qwen-VL, and diffusion models.
 </p>